Research map / Label-efficient and robust learning / Recognition and long-tail
Poly-View Contrastive Learning
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Proposes poly-view contrastive learning objectives for tasks with more than two related views, derived from information maximization and sufficient statistics.
Problem
Standard contrastive learning matches pairs of related views against unrelated negatives; the paper targets settings with more than two related views (poly-view tasks), where the usual pairwise objectives may not exploit all related views.
Approach
Derives new representation learning objectives for poly-view matching using information maximization and sufficient statistics, and argues that with unlimited computation the approach yields sufficient representations for all views.
Experimental setup
Builds on: contrastive learning InfoNCE
Closest papers by meaning
- What Makes for Good Views for Contrastive Learning? (NeurIPS 2020 · similarity 0.88)
- Conditional Negative Sampling for Contrastive Learning of Visual Representations (ICLR 2021 · similarity 0.88)
- A Simple Framework for Contrastive Learning of Visual Representations (ICML 2020 · similarity 0.88)
- $\mathbb{X}$-Sample Contrastive Loss: Improving Contrastive Learning with Sample Similarity Graphs (ICLR 2025 · similarity 0.87)
- On the Similarities of Embeddings in Contrastive Learning (ICML 2025 · similarity 0.87)
- FFF: Fixing Flawed Foundations in Contrastive Pre-Training Results in Very Strong Vision-Language Models (CVPR 2024 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.